GAMSSA A new Global Australian MultiSensor SST Analysis

1
GAMSSA A new Global Australian Multi-Sensor SST
Analysis

• Helen Beggs
• Centre for Australian Weather and Climate
  Research,
• Bureau of Meteorology, Melbourne, Australia
• Presented at the 9th GHRSST-PP Science Team
  Meeting, Perros-Guirec, France, 10 June 2008

2
Overview

• Description of new test Global Australian
  Multi-Sensor SST Analysis (GAMSSA) system
• Assessment of Test Global Australian Multi-Sensor
  SST Analysis (GAMSSA v1.0) system
• Comparisons of GAMSSA with other SST analyses
• Comparisons of input satellite data streams with
  buoy observations

3
BOM Global Australian Multi-Sensor SST Analysis
System

• V1.0 Alpha testing started 6 Dec 2007
• Beta testing started 4 May 2008
• Depth Foundation
• Resolution Daily, 1/4
• Domain global
• Based on RAMSSA v1.1 code
• Observation correlation length scale 20 km
• Background correlation length scale 50 km
• BGF Combination of previous days GAMSSA and
  Reynolds climatology
• Data Inputs
• 9 km NAVOCEANO GAC AVHRR (NOAA-17, -18, METOP-A)
  L2P
• 25 km AMSR-E (Aqua) L2P
• 1/6 AATSR (EnviSat)
• Buoy and ship obs (GTS)
• 1/12 NCEP ice edge analyses

4
Performance Analysis 1 Oct 2006 31 Mar 2008,
60E to 170W, 20N to 70S

• Weekly 1 Operational SST1m
• Daily 1/12 RAMSSA v1.1 SSTfnd
• Daily 1/4 GAMSSA Alpha Test SSTfnd

5
How does GAMSSA v1.0 compare with other daily,
multi-sensor (MW IR) SST analyses?
6
BoM 1/4 GAMSSA v1.0 SSTfnd
BoM 1/12 RAMSSA v1.1 SSTfnd
OSTIA 1/20 SSTfnd
Reynolds 1/4 AMSR-EAVHRR SST1m
7
BoM 1/4 GAMSSA v1.0 SSTfnd
BoM 1/12 RAMSSA v1.1 SSTfnd
ODYSSEA 1/10 SSTfnd
GMPE 1/2 SSTblend
8
GMPE Daily 1/2? SSTblend
ODYSSEA Daily 1/10? SSTfnd
Reynolds Daily 1/4? AMSR-EAVHRR SSTblend
OSTIA Daily 1/20? SSTfnd
9
GAMSSA v1.0 SSTfnd Analysis minus Other SST
Analyses

• Over region 180W to 180E , 75N - 75S
  (global)
• Period 1 Jan 31 Mar 2008
• GAMSSA BoM Operational Weekly 1 SST1m 0.0
  0.6C
• GAMSSA GMPE Daily 1/2 SSTblend 0.1 0.4C
• GAMSSA Reynolds MWIR Daily 1/4 SSTblend 0.0
  0.6C
• GAMSSA ODYSSEA MWIR Daily 1/10 SSTfnd 0.0
  0.4C
• GAMSSA RAMSSA MWIR Daily 1/12 SSTfnd 0.0
  0.3C
• GAMSSA OSTIA MWIR Daily 1/20 SSTfnd 0.1
  0.4C

Mean(GAMSSA Anal)
Std Dev(GAMSSA Anal)
10
Analysis SST(date) minus Buoy SSTfnd(date1)

• Over region -180E to 180E, 70N to 70S
  (Global domain)
• Period 1 Jan 31 Mar 2008
• GMPE Daily 1/2 SSTblend -0.04 0.61C
• Reynolds MWIR Daily 1/4 SSTblend 0.01
  0.65C
• ODYSSEA MWIR Daily 1/10 SSTfnd 0.02 0.55C
• GAMSSA Daily 1/4 SSTfnd 0.01 0.52C
• OSTIA MWIR Daily 1/20 SSTfnd -0.04 0.45C

11
Analysis SST(date) minus Buoy SSTfnd(date1)

• Over region 60E - 180E , 20N - 65S (RAMSSA
  domain)
• Period 1 Jan 31 Mar 2008
• GMPE Daily 1/2 SSTblend -0.10 0.52C
• Reynolds MWIR Daily 1/4 SSTblend -0.04
  0.59C
• ODYSSEA MWIR Daily 1/10 SSTfnd 0.11 0.50C
• GAMSSA Daily 1/4 SSTfnd -0.03 0.44C
• RAMSSA MWIR Daily 1/12 SSTfnd 0.04 0.43C
• OSTIA MWIR Daily 1/20 SSTfnd -0.04 0.40C

12
Input Satellite SSTfnd Statistics 60E to
170W, 20N to 70S

• Shows residual bias in SSTfnd estimates from
  NOAA-18 and AMSR-E on Aqua (equator crossing time
  of 0130 LT) due to diurnal warming
• Difference between HRPT and GAC AVHRR SSTfnd
  biases (from same sensor on same satellite)
  indicates impact of the L2 SST processing method

13
Input Satellite SSTfnd Statistics 60E to
170W, 20N to 70S

• Difference between NESDIS and NAVOCEANO GAC AVHRR
  SSTfnd error (from same sensor on same satellite)
  indicates impact of applying the bias per pixel
  (SSES) corrections from the NAVOCEANO GHRSST-PP
  L2P files.
• HRPT AVHRR SST is NOT blended in GAMSSA so least
  accurate input to GAMSSA is AMSR-E with 0.5?C
  St. Dev. Error and greatest number of data points.

14
Conclusions

• Over 7 Jan 31 Mar 2008 GAMSSA v1.0 1/4 SSTfnd
  Analyses
• Have less RMS error than BoM Weekly 1? SST
  analysis over 60?E to 170?W, 20?N to 70?S
• rms(Obs BGF) of 0.5C c.f. 0.8C
• Agree more closely with ODYSSEA SSTfnd analyses
  than with GMPE SSTblend, Reynolds SSTblend or
  OSTIA SSTfnd analyses globally
• GAMSSA ODYSSEA 0.0 0.4C
• GAMSSA(date) Buoy SST(date1) 0.01 0.52C
• ODYSSEA(date) Buoy SST(date1) 0.02 0.55C
• Expect to become operational BoM product by end
  of 2008
• Can the GMPE Project use a Beta Test GAMSSA?

15
(No Transcript)
16
BoM Regional Australian Multi-Sensor SST Analysis
System

• V1.0 Operational 13 Jun 2007
• -gt RT input to BoM NWP models
• V1.1 Operational 26 Oct 2007
• Depth Foundation
• Resolution Daily, 1/12
• Domain 60E - 170W , 20N - 70S
• Observation correlation length scale 12 km
• Background correlation length scale 20 km
• Based on legacy BoM optimal interpolation
  regional SST analysis system (Smith et al., 1999)
  
• Data Inputs
• 1 km HRPT AVHRR (NOAA-17, -18)
• 9 km NESDIS GAC AVHRR (NOAA-17, -18, METOP-A)
• 25 km AMSR-E (Aqua) L2P
• 1/6 AATSR (EnviSat)
• Buoy and ship obs (GTS)
• 1/12 NCEP ice edge analyses

Daily foundation SST analyses available by 0300
UT as netCDF L4 files from http//godae.bom.gov.au
and http//ghrsst-pp.jpl.nasa.gov
17
Relative Weighting of Data Streams in OI Analysis

• OI system requires an estimate of standard
  deviation (?tot) for each input observation
• SD error ? relative weight to give each
  observation in the OI analysis relative to each
  other and the background field
• Total variance ? instrument var
  representativeness var
• ?tot2 ? ?instrument2
  ?space2 ?time2
• BLUElink analysis system compares each satellite
  SSTfnd data stream over a 24 hour period with any
  buoy SSTfnd observation within half the satellite
  sensor resolution to determine total observation
  standard deviation error.

18
Data Inputs to L4 SST products over Australian
region
19
Drifting and Moored Buoys (GTS)
9 km x 4 km GAC AVHRR (IR)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)